Spontane Ideen, die ich zwischen 2013 und 2020 hatte

Seit 2013 habe ich einige meiner spontanen Ideen in einer Datei namens „idei spontane“ auf meinem Desktop abgelegt. Ich hatte vor, diese Ideen selbst umzusetzen, aber da ich ab März 2021 als PostDoc zu arbeiten beginne, denke ich, dass ich nicht in der Lage sein werde, diese Ideen zeitnah zu implementieren. Aus diesem Grund, und vor allem, weil ich glaube, dass diese Ideen für den einen oder anderen zukünftigen Studenten oder Entwickler von Interesse sein können, werde ich sie in diesem Blogpost veröffentlichen. Die spontanen Ideen, die ich zwischen 2013 und 2020 hatte, sind:

• To create a game for children, cute, in which a small animal or bird is angry with the sun. The reason for the upset is that his friend, the snowman, melted, and he was attached. The denouement is his realization that there are seasons and a cycle in nature, he understanding this and eventually becoming friends with the sun. In another edition of the game, he has as help the sun, who is now his friend, with whom he goes in search of the moon, but which he does not find (what he does not know, is that in the presence of the sun, he will never see the moon in its splendor). The denouement will be, that he will have to say goodbye to the sun, to make room for the night to settle
• The idea of phone application: When I get to sleep, the ones who call me, get an automatic message, that I’m sleeping and I’ll call them when I wake up. (I think there is such a thing, but I do not have time to check now – noted here to remind me).
• Making an option/camera to do the following: To enter in the photo and the photographer automatically (if you want this, of course, having the mode activated at the desire), captured being in the same second as the others who are taken the picture. When he presses the button to capture, he will be taken a picture with the front camera and him, and the software will place him next to the others. His position can be defined: either to appear in the middle of them or at the edge. With each update of the software, you will be able to modify several elements. This will be the first time a friend will not have to be left on the sidelines when no one is around to take your picture on vacations, etc.
• dogs are good, very close, and dedicated largely because they were taken from their mother as small and sold, like dogs of the breed, etc., and that feeling that they lack mother, that lack of affection and reflects it on the man who is friendly with him and cute. And if the master happens to be aggressive with him, he endures and gets over it, because he thinks that trust is more important than what he feels at the moment, having a sense of duty towards the one who feeds him, shelters sometimes plays or punishes him. Dogs cannot sit on the throne on the same level as people because people do not notice it, they are occupied largely because of the ego, with the search for a throne on which to ascend as kings and praise their position, they want to be admired, they have inferiority complexes; I refer here to those who do not have meditation as an intrinsic part of personal lifestyle.
• An idea on 23.05.2014: already the world is starting to get bored on Facebook often. to change this and bring back the interest to interact with each other, it would be great for some users to offer facebook a certain badge, a certain „surprise gift puzzle badge“ What does that mean? many people from the same continent or another, it is questionable the idea, to have to meet personally in order to be able to complete the image of a gift, and the gift to be even greater the distance between users. For example, if the gifts are distributed among a few Europeans, they receive a trip to Asia as a gift together and food, drink free accommodation. If it’s between users from the same country, get a weekend trip with a barbecue, music, etc. in their country. You can work on the idea. Possibly, another idea: the friends you have online in your list to receive a puzzle badge and you start to talk with friends and ask them, what piece has / with which you can form the image of the gift ;d e e.g. if you form with someone an exit to ice cream and cake. You can print the image formed as it has a qr code say.. and pay with that, which will be scanned by the store that is a Facebook partner. it would be maybe more ok that for those singles, the missing pieces in the image to always be to people of the opposite sex, to be able to be a nicer moment, to open between them, because the planet needs more open people, more harmonious.
• business card made of modern material that digitally displays any change that takes place in the company’s domicile, email address, name, picture of the person appearing on the business card, etc. to reset when the employee dies or is fired, or the company dissolves, etc.
• An idea on July 10, 2014: when in a violin concert e.g. played after musical notes, usually the score returns manually. Can’t you use an iPad device? Or rather a digital paper that displays the musical notes that follow automatically, the violin being connected by a device with its own software. Of course, you can improve the software, e.g. when you finished the page, to show you where you made small rhythm mistakes, etc. Also when you sing, have sensors and know where you left off. I mean, not to display the notes only after time, but as you sing them, the ones sang to have one color, the ones that come, another.
• Creating a Facebook app: I enter a chat room, and the chat is anonymous. I can be asked questions from all Facebook users, or just my friends, having the freedom to choose privacy settings. Then in the chat, I will be asked more direct and even intimate questions from people who want an answer and did not have the courage to ask me directly, so as not to show their identity. The answer to the questions will always be given as honestly as possible after everyone’s education. If the answer is satisfactory and welcomed by the questioner, he can vote to become visible who he is, to show his identity, etc.
• 22.07.2015: Today while I was shaving I came up with an idea to create a razor sharpener while the razor was on stand-by. The Gillette razor, for example, has that head blade that needs to be changed every 1-2 months, depending on how often you shave. Well, we’re changing that „head“ just because it’s got a blunt blade and it doesn’t cut well anymore, right? What if instead of that plastic cap in which the machine sits for so long closed until the next barbering, it actually had some small fine pebbles that once rubbed with the blades to sharpen them, thus being able to double or triple the life of that blade? Or possibly to create a band like a shell but with small granules like „șmirglu“, which is already perfectly designed so that you can pull it suddenly through all 5 blades. to sharpen them (Like the seam that is sharpened by the mower on the field rubbing with a stone the cut of the seam and thus sharpening it.)
• 21.09.2017: Programming languages require an extra effort to be learned by those who do not have a penchant for them. My idea is to create a program using AI (neural networks, machine learning, deep learning) that writes in a programming language (like python, c++, PHP, etc.) the requirements by voice (or text) of the user, a kind of Siri, but for programming and not only to display that code but to be able to run it. For example, a student has the homework to make a program in python that displays a frame with menus File, Edit, etc., and contains a text and a picture. Well, he without knowledge, could simply tell the program requirements, and the program could identify the keywords and realize, e.g. deep learning algorithms, that theme.
• 15.10.2017: Reading an article about DNN, and knowing that those approx. 86 billion neurons in the human brain are connected to each other by synapses (weights) between dendrites which are elements of the neuron that insert the signal/information into the neuron, and axon, which is an element of the neuron that receives the signal generated by the neuron after receiving it through dendrites, and which sends it forward through synapses to other dendrites (hence other dendrites), came to me the following idea: Can these neurons be stimulated to learn through frequent? It is said that classical music, e.g. Mozart, Beethoven, etc., has been shown to stimulate the brain. As for all the talk, I also know about meditation, which says it radically changes your gray matter :)) … Schuman Resonance has been shown to be the Schuman Resonance (see video https://www.youtube.com/watch?v=FttrOK1ec4Q&feature=youtu.be) or the frequency/pulse of our 7.83 Hz planet is identical to that of the human brain and that the pulse of life is the same. In order not to lose too much in words, my question is: Can we develop DNN algorithms that accelerate the flow of information between neurons using sacred geometry, like golden ratio? They say Mozart and others used it in their music, hence the peace when you listen to it.
• I see in many articles that they talk about a large number of neurons and data make an algorithm better, the same I hear that a man (or animal, elephant type) who has a larger brain has greater capacities. But what if a large number of neurons isn’t the right track? I remember Nassim Haramein (see video https://www.youtube.com/watch?v=peYrSP8cke8) who took to the fun those with Hadron Collider who were trying to discover god’s particle, haha, and he said that it is not possible, because it is not important to dissect, because you will dissect forever, and he came up with a scientific article describing a geometric paternal of existence, the flower of life, tetragrammaton, etc.
  Maybe it’s not the number of neurons but the way they’re arranged that’s important, which is the pattern .) Also, see Vortex Based Mathematics. But what role does the stomach play in human intelligence? Especially the point of grace? Why in the definition of a neural network we say we want to mimic how the human brain works and talk about neurons, etc. But doesn’t the stomach have a special place too? And Osho was talking about the navel area.
• Idea 19.12.2017: Create a tool to search for images on Facebook using the name or the standard numeric code generated by Facebook in the name of the photo when the user saves the photo. For example, I see a very beautiful photo of Bucovina on my wall, I click on it to enlarge it, and then right-click, save the picture, then I introduce a new name or leave this standard given by Facebook, which is actually a unique code, then I click on save to have the image saved to my computer.
  Well, I realized that if I wanted to post the image on my blog and quote the source, I cannot remember which page is taken, so it would be very useful a tool was to enter the name of the image, that is exactly that unique code generated, and then I can find that image on Facebook having that code. Of course, probably many have downloaded and reloaded the same image and you will say that I will probably find the image countless times, but it is not, because every time someone reloads an image on Facebook, even if it is the same image, in reality, will have a unique numeric identification code on Facebook.
• Idea 28.02.2018: After I was angry that there are so many „fines“ penalties in Germany if you do not follow the law, I came up with an idea of an application or system based on deep learning to do the following things:
  – To remind you of what you need to be totally legal.
  -An example would be to always have the vest in the car if you go on the road, do not forget to check the oil, TÜV, 1.6mm depth on the wheels, etc.
  – in case you risk getting into an altercation, read your pulse and remind you to keep calm because the law says don’t hurt, etc.
  – the idea would be especially successful in Germany where in general any minor non-compliance is punishable by a police check.
  Another option would be: not being able to leave the house if you forgot to get your ID that legally you must have with you all the time, or you can not start the car until you solve the problems displayed, etc.
• 04.04.2018: When I refer to a scientific article, the order of references in my paper does not begin with 1, 2, 3 … etc., but random, depending on how I find something interesting to add, modified (98, 4, 14, etc.). The idea is to create a scientific article to do a study of scientific articles from the last 5 years and graphically show the patterns in which the references of researchers were added. Perhaps it is possible to demonstrate how creativity in the style of adding references manifests itself as well as the importance of some articles compared to others.
• 27.08.2018: Flyers placed in the post by the pizza siders have a bad smell because they are usually freshly printed and still persist the smell of „ink“ (similar to the smell of new books). The idea is to replace this smell with a pizza smell in the case of pizzeria flyers. There must be a possibility to change that smell from the ink production phase. In this way, the flyers will smell like the product to which it is advertised.
• 02.09.2018: the idea of 3 articles:
  1) Detection and Classification of Brancusi Sculptures using CNNs
  2) Detection of Romanian Classical (or traditional) Music using RNNs
  3) Detection of Eminescus’s Poetry (or Romanian poets) using RNNs
• Article idea: Similar to Ray Tracing technology in RTX GPUs, is there the possibility of generating light as artificial solar radiation using AI? Thus a solar tracker would capture energy non-stop.
• Another good idea: Regionalism Voice Detection and Classification of Moldovans, Olteans, Banateans, Ardeleans using RNN (we train on several audio files found on youtube e.g. by categories: Moldovan voice, Arlene, Bananas, Olteni, etc. and we classify them by region where they come from and possibly by sex – the idea came today 7 Sept 2018) + apply voice to text (it can have many applications like grandma telling stories, a chat is output as the voice of the person is typing, can have huge commercial success.)
• capture of solar energy through mirrors or focus through a magnifying glass
• creating an algorithm that generates from a SINGLE 3D IMAGE, 100,000 images, having the object of the class in question, displayed in different poses.
• Sharpee speculated that feeding inputs with hyperbolic structure into machine learning algorithms could help with the classification of less-structured objects. „There’s a starting assumption in deep learning that the inputs should be done in a Euclid metric,“ she said. „I would argue that one could try changing that metric to a hyperbolic one.“ Perhaps such a structure could further optimize deep learning systems. https://www.wired.com/story/artificial-intelligence-has-a-strange-new-muse-our-sense-of-smell/?_lrsc=8824cbaa-66ca-4bd8-ab5c-f09584b422ae
• Idea 4oct2018: creating an application that shows me automatically when I enter a folder with images belonging to different classes, which classes are detected, and to „mask, like maskRCNN“ to make it much easier to select the unwanted ones. Thus I would help an entire community of data scientists, who, especially in computer vision, must clean the dataset of images (e.g. in a folder with images containing dogs, there could also be images of people, etc. … when the Imagenet dataset was created, they used Amazon Mechanical Turks to clean the images and give them the label manually. Well, I would like to no longer need to select them manually one at a time, but to be selected with over 90% accuracy automatically by a deep learning model. Later edit: I did something in this direction with my Deep Learning-Based Computer Vision Application with Multiple Built-In Data Science-Oriented Capabilities.
• use neural networks to detect and identify objects by their shadows only -idea from 31oct2018
• detection of animals by a drone, showing the location of the animals and what kind of animal; the idea is also good for wildlife photographers that usually wait to pose a particular animal, which can now know in advance whether the favorite animal is in the area or not and where exactly is as a cardinal point.
• an article about the order in which the references were quoted in a scientific article. To show over a period of 10 years e.g. that the number of citations is higher on articles that have references quoted in ascending order or those in chaotic order. A comparison will be made between the fields of research, showing e.g. that articles with references in chaotic order in biology are more cited than those in ascending order from computers -idea 4nov2018
• detection of a class (humans, animals, birds, etc.) by training images that represent their shadow. Of course, due to the position of the sun, the shadow will have different angles, but I’m sure it will be possible to perform tests much faster using artificial light in a room. Or more than that: creating a simulator that casts light from all possible angles on an object (class) and automatically saves pictures of those shadows from all angles, for a sun-like rotation from sunrise to sunset.
• creating a website where people post the most beautiful pictures in Romania. Then, take the coordinates of these pictures with the location, time, month, year, etc. exactly where the picture was taken and created a map where everyone can see exactly the place on the map where the photo was taken with offers of accommodation and food in the area so that the world can enjoy the landscape directly. With such an idea, everyone will be able to go and photograph and experience the „secret“ location of the original photographer.
• An article that presents the best scientific articles not only by the number of citations but by the impact they have, because „We measure the impact in terms of the number of citations, often ignoring that many of these very mentions use one’s work as a baseline to show how it is no longer competitive.“
• creating a solution to fix a presenter’s gaze when reading a text on the teleprompter (often we realize when YouTubers prefer to read a text while presenting a certain theme, e.g. Linus tech tips techquickie)
• Create a team in ACSA to create hands-on products (e.g. an automatic screwdriver using Arduino, the most efficient solar tracker in Romania, mobile applications for detection of iii, animals, birds, etc.). Also, a site with the team where you can display the results of research of great national importance and where these projects can be displayed for the wide publication as well as for sale, the money being used for future successful projects.
• Idea 30.01.2019: BookMe – a website where you can see my profile at the top, and an enormous calendar at the bottom. This calendar will be able to scroll so that anyone can see my days off or busy (occupied, in the sense that they are booked and therefore can no longer be booked by them). BookMe can be a success because: Maybe a friend would like to make a reservation to give my whole day only to him when I enter the first day of retirement :)) On the site as a user, I will be able to create some events that they can book. For example, they’ll see that in 1 or 2 years I’ll be going to Iran for the first time, and they really want to come with me, spend the day with me.
• Idea 13.03.2019: Create a website where people can select the Land, the City, the preferred Zahnarzt, and then fill out the „Zahnarzt Anmeldeformular Neupatienten“ directly online, without having to download it as .pdf, as is the procedure all the time in Germany. The website could contain all the Zahnarztpraxis in Germany, possibly charging a very minimal fee to the praxis, otherwise, to be free, and money to be possibly from the paid advertisement (having a large volume of potential patients, some clinics would like to appear on the front page to attract new clients).
• Idea 28.03.2019: Create a device that can read the contents of the letters even if they are closed. I think it is possible to implement a technique (laser genre) that extracts/identifies the exact position of the letters. After all, a handwritten or printed sheet has all the cellulose as a background. Only „what leaves traces“ on it is either graphene in the case of pencil, or ink, or paste, or ink from the printer, etc. Once the content sheet is separated (letters, numbers, etc.), you can make a map of them on a digital A4 sheet, as a kind of digital xerox/scan.
• Idea 22.04.2019: Sending messages using chat, be it just private, can often be a hard thing to do, especially if the jealous girlfriend is there and you want to send a warm thought to a friend from the past. The idea is to create a mechanism to be able to convey and receive feelings.
• Idea 07.06.2019: The fact that the architecture and functionality of airplanes are bad compared to the flight of birds and insects, the idea is to combine the anatomical knowledge of insects e.g. with those of mechanical architecture, etc. and create exact prototypes after the copy of the insect to see if you can replicate flight objects much more advanced than the current ones that are pathetic in case of malfunctions (for example, why the wings of current airplanes are so small compared to their body, plus they look like wings without feathers when the wings of birds are actually 2-4 times larger than their body, plus they have feathers?). To do: Scanned an insect in 3D and 3D printed all the elements of the body, then tried to make flying objects even for ordinary people (Yes, the world will be able to fly! plus they will not die innocently in case of fires in tall buildings or in case of a possible plane accident)
• Idea 14.07.2019: Usually when training a CNN with images as the input they use CONV filters to extract the most important features and reduce the size of the images. It’s the same with pooling layers.
  My idea is: I want to „weigh“ every pixel in an image, then create an algorithm that weighs groups of pixels, resulting in the hardest ones actually representing the most important features in that image
• Idea 01.08.2019: Like the idea above, only here „measure“ each pixel from the point of view of energy consumption, showing that a certain value of a pixel = a certain amount of energy consumed using an X algorithm. This way I will be able to predict almost exactly by the pixel value, how much energy is consumed, or how much energy we need even before we start training or inference.
• If birds left tracks in the sky, what would they look like? For years Barcelona-based photographer Xavi Bou has been fascinated by this question, but of course, birds in flight leave no trace—at least none visible to the naked eye. Bou had a plan though and he has spent the past five years trying to capture the elusive contours drawn by birds in motion, or, as he says, “to make visible the invisible.” He began exploring photographic techniques that would allow him to show the beauty of birds in a way not seen before and ultimately he chose to work with a video camera, from which he extracts high-resolution photographs. After he films the birds in motion, Bou selects a section of the footage and layers the individual frames into one image. He finds the process similar to developing film: He can’t tell in advance what the final result will be, but there’s one magical second, he says when the image—chimerical and surreal—begins to emerge. Before Bou began this project, which he calls “Ornitografías,” he earned degrees in geology and photography in Barcelona, then worked as a lighting technician in the fashion industry. This current work, he says, combines his passion and his profession. “It’s technical, challenging, artistic, and natural. It’s the connection between photography and nature that I was looking for. Ornitografias is the balance between art and science, a project of naturalist discovery, and, at the same time, an exercise of visual poetry.” Text from: https://www.facebook.com/cineandreea/posts/10156338316671851
• Idea 06.11.2019: Training a neural network to detect areas where forest cuttings happen (detection of trucks among trees, many fallen trees, group of people in the area), the model is later implemented on a drone. At the same time, when illegal logging is detected, an alert in the form of a text message with the exact location (similar to https://towardsdatascience.com/average-rolling-based-real-time-calamity-detection-using-deep-learning-ae51a2ffd8d2) will be sent to the ranger who takes care of the forest in order to be able to make quick decisions and act in real-time. Later edit: even detecting traces of large machinery in protected areas, could be from those who are there illegally. Possibly made dataset and with traces, and with various machines that are usually used in forests.
• Idea 22.04.2020: Usually, when we want to enter the apartment we have a lot of keys in hand and we have to look for the right key for a few good seconds sometimes, especially if we don’t have enough light. The idea is to put a few mm LED on each key, and when the key is within 10cm of the doorknob/door lock because inside the house we have mounted a very small device that gives magnetic energy only to that key that will fit to open the door, that key will light up thanks to the LED. Thus, we decrease precious time, especially in case of danger, fires, accidents, emergencies, etc. This idea will not require the installation of a biometric solution, which is expensive, but will be able to operate on the current installation in almost any home in the world. Ps. One could also get an idea in the future with a system in which the language of the door to change shape and return to normal form only when the near key to 10cm has an SHA-256 algorithm that gives it a hash as authorization, and she knows that it is the user in question, but that’s another idea.
• Idea 15.06.2020: Create a Deep Learning application to correct a user who speaks a foreign language using voice (possibly text and speech). Let’s take german as an example: at the moment many of the beginners are corrected by other people more experienced when they say e.g. „Ich habe mich geduschen“ with „Ich habe mich geduscht“. These experienced people are not always around beginners, which leads to this idea of an application that has the potential of being commercially successful. The initial version will have one or several foreign languages (trained models), but in future versions will have „all“ international languages. It will also be possible to create „background noise removal“ options for the application to hear clearly what words the user uses as well as options for correcting to be done in the form of voice or text (especially when the user is in public transport), or both at the same time.
• Idea 25.06.2020: A being needs at least air, water, food, and heat to survive, air and water being the most important. The air is naturally inspired and exhaled by the body, but the water does not, it must be found. Today’s idea is the creation of a bioinspired device that produces water „out of thin air“ even if very little per hour, and that is possibly mounted in the mouth, either in humans or in animals. I think it’s possible because oxygen being O2 and H2O water, this device could somehow only produce H, and when it is combined in the oral cavity with the O2 that is inspired, it could produce at least a drop of water per minute. In later editions would produce more than that, even it could be set to produce a desired amount by the user or be bought directly with a predefined water output. Such a device could be successful because there are many people who could have water even in their sleep not being set when they wake up (especially old people who need a glass of water at their head), in tropical areas, deserts, missions, etc.
• Idea 02.07.2020: Create sensors that will be attached to the human body and that charge with sunlight or air and communicate via GPS or 4G or 5G. The idea is to increase the mobility and safety of a person when he is in a state of emergency and does not have a mobile phone with him. For example, if a man wants a taxi in his location, he will press/activate the specific sensor. If he is in danger, he will press/activate that sensor that will send signals that he is in danger (in case of kidnappings, lost in the forest, disasters, etc.) of a police department, they know for sure that it is a danger as well as the exact area where the person is so that he can intervene immediately.
• Idea 31.07.2020: The first idea is: Create a device that will help any person write, draw like a professional, without knowing or having talent necessarily. The person will only have to hold in his hand either a pencil designed specially or to wear a glove, and with the help of powerful magnets will create a magnetic field that will move/move the pencil/hand where necessary to accomplish the task, similar to the movement of those 3D printers after a previous design. Thus, many could make pencil sketches on paper very very cheaply, because the program (possibly using Deep Learning to know the exact position, increases speed and quality, etc.) uses the device that works with magnetic forces, without the need for „special 3d ink material“, requiring only a simple pencil (pix, pen, carioca, etc.) cheap. The second idea is: creating a hardware design (e.g. in Verilog) that has a function that gives priority to bits that „are loyal customers of a bar and buy daily alcohol after work“, even when „they happen to have no money on them“ because the function knows that they are bits that come daily, so they are „loyal customers who can receive „access“ even on debt“. This could create a metric or something where, in case of errors, they will be allowed to pass, because, in the next round, the bits will „correct the damage done /pay the debt“. Even if these phrases don’t sound good, they deserve a discussion for a potential scientific article.
• Idea 19.11.2020: Creating an application for mobile phone or even hologram glasses, to help the customer of a Supermarket find the product searched instantly, without having to walk in many rows or look for various shelves. For example, sometimes I want to buy „poppy seeds“ but I don’t know exactly where to find them. Even if in a supermarket the areas are called „Drinks“ or „Dairy“, etc., they do not tell or show me exactly where that product is placed on the shelf. This application can be very helpful, especially when the supermarket is full of customers and you are in a hurry.

Bitte entschuldigen Sie meine grammatikalischen Fehler im Text, aber weil meine ursprünglichen Ideen in rumänischer Sprache geschrieben wurden, um Zeit zu sparen, habe ich nur Bing Translator für die Übersetzung des Textes aus dem Rumänischen in die englische Sprache verwendet. Übrigens, Bing Translator ist besser als Google Translator, was die rumänische Sprache betrifft. Ich freue mich auf eine rumänische Übersetzung, die von DeepL (hoffentlich bald) angeboten wird.